JPS63178300A - Voice encoder - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a speech encoder for efficiently encoding speech signals.

2. Description of the Related Art Conventionally, PCM, ADl'CM, and the like have been developed and put into practical use in order to efficiently encode voice, especially voice in the telephone band. However, the demand for efficiency has increased further with the start of digital line services. Currently, multi-pulse systems, APC systems, and the like have been proposed as audio encoders at a bit rate of about 16 BPS.

Problems to be Solved by the Invention The multi-pulse method described above uses a plurality of pulses as the sound source signal of the synthesis filter.

The pulse height and position are sent as sound source information. On the other hand, AP
In the C method, waveform information (values of all samples) is sent as the sound source signal of the synthesis filter.

′ Among these methods, the multi-method uses pulse information, so it is efficient when the sound source pulse is large, but when the sound source pulse is small, such as in the case of fricatives, it becomes like continuous noise, so the number of pulses is low. There is a drawback that there is not enough signal, and the signal-to-noise ratio deteriorates.

On the other hand, with the APC method, there is no problem with continuous noise-like waveforms, and when there are large pulses (voiced sounds), information bits are assigned to all samples, so pulse information is cannot allocate bits to S
There is a drawback that the N ratio becomes worse.

The present invention has been made in order to eliminate the above-mentioned disadvantages inherent in the conventional technology, and therefore, an object of the present invention is to efficiently encode audio signals and ensure a good S/N ratio. The object of the present invention is to provide a novel speech encoder that achieves the following.

Means for Solving the Problems In order to achieve the above object, the speech encoder according to the present invention comprises:
It has both a multi-pulse system and an APC system, and also has a voiced/unvoiced determiner, and a switch that selects the output of the multi-pulse system when voiced and the output of APC system when unvoiced.

Embodiment Next, a preferred embodiment of the present invention will be specifically explained with reference to the drawings.

FIG. 1 shows an embodiment of the present invention. +00 is the input terminal,
lO is a multi-pulse extractor (sound source pulse extractor), 20
is the residual signal waveform extractor (sound source waveform extractor).

30 is a characteristic parameter extractor (characteristic parameter analyzer)
, 40 is a switching controller (voiced/unvoiced determiner), and 50 is a switch.

From the input voice, a multi-pulse extractor IO extracts a multi-pulse (W source pulse) based on the output of the characteristic parameter extractor 30, and a residual signal extractor 20 extracts a residual waveform (sound source waveform). Both are controlled by a switching controller 40 and a switch 50 so that a multi-pulse (sound source pulse) is output when voiced and a residual waveform (sound source waveform) is output when unvoiced. The output of the Sui-Sochi 50 (sound source information), the output of the characteristic parameter extractor 30 (characteristic parameter), and the output of the voiced/unvoiced determiner (switching controller 40) are combined by a multiplexer 61) and output to the speech encoder. ′ becomes the output.

FIG. 2 is a block diagram showing an example of the switching controller 40. As shown in FIG. Insert the sound source from the multi-pulse input terminal ttst,
Synthesis filter 4 defined by characteristic parameter input 102
A comparator 43 determines whether the difference from the original sound (audio input) 103 is larger than a certain level (T11) and a control output 104 is obtained.

As described in detail, by using the speech encoder of the present invention, the multi-pulse system can be used for voiced sounds where the pulse sound source is more efficient, and the modem and blind voices where the residual signal waveform is more efficient. Since the APC method using the residual waveform is selected for the signal, it is possible to compensate for the drawbacks of both methods.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention. 100...Input terminal, 200...Output terminal, 10.
・・Multi-pulse extractor (sound source pulse extractor), 20・
...Residual signal waveform extractor (sound source waveform extractor), 30...
・Characteristic parameter extractor (characteristic parameter analyzer), 4
0...Switching controller, 50...Switch, 60...
Multiplexer FIG. 2 is a block diagram showing an example of a switching controller used in the present invention.

Claims (1)

[Claims] an analysis means for obtaining characteristic parameters and residual power of speech using a linear prediction method; a sound source pulse extraction means for outputting a prediction residual as pulse height and position information using the characteristic parameters; a sound source waveform extraction means for outputting the prediction residual as waveform information using the sound source waveform extraction means;
a switch that inputs the output of the sound source pulse extraction means and the sound source waveform extraction means and outputs one of them; and a switch that inputs the output of the analysis means and the output of the sound source pulse extraction means. 1. A speech encoder comprising: a switching controller for controlling a switching controller;